Transformation

Data in FACS data files is usually stored as already log-transformed data, or as linear raw data (see also in FACS data. Fluorescence data is usually shown on a logarithmic scale. Hence, the linear data needs to transformed for the plots. FCSalyzer offers the logicle transformation developed by David R. Parks, Mario Roederer, and Wayne A. Moore - for more, please see below.

The dialog "File - Format Datafiles" has a tab to view and change the transformations for the parameters of the open FACS data files. There are three types of data display: linear, log and logicle. The many different logicle transformations display the negative values to a different extent - from 0% of the scale to about 30% of the scale. Values that are below this setting will be shown on the axis. Simply try different setting until you get a feeling for them.

There are a few things to keep in mind:

• Changing the transformation only matters for displaying the events. The actual values and therefore the statistical values are not altered.
• Changing the transformation of an parameter of a file will apply to all plots showing the file in a document.
• Regions are not adapted when a transformation is changed. Make sure to re-check that all regions are set correctly!
• In principle, all data files can be shown on linear, log or logicle scale. However, only linear float data is really suited for all transformations. Classical FCS2 files often use channel data that is already in log scale. Changing the transformation here might not give good results.
• Data values that are too low or too high for the applied transformation are shown as the lowest/highest possible value. For example, when using the log-Transformation, values of 1 or lower are all shown in transformed channel 0. But this is only for displaying the data - the calculation data remains unchanged.

And now a few words about the logicle transformation:

Displaying linear data on a logarithmic scale sounds easier than it is done. There are two problems: for one, log-transformation of low values leads to single, separate lines ("fencing"). The bigger problem is, that there is no logarithm of zero and values below zero. Such values can come up for example due to compensation.

Fortunately a transformation was developed and published by David R. Parks, Mario Roederer, and Wayne A. Moore ("A New 'Logicle' Display Method Avoids Deceptive Effects of Logarithmic Scaling for Low Signals and Compensated Data"), followed by an update by Wayne A. Moore and David R. Parks ("Update for the logicle data scale including operational code implementations"). I do not understand a word in these papers. But the latter has supplementing information that provides a JAVA library for the logicle transformation. That saved my day. I am very thankful to the Stanford University for providing this library and generously stating that: "The Logicle method is patented under United States Patent 6,954,722. However, Stanford University does not enforce the patent for non-profit academic purposes or for commercial use in the field of flow cytometry."

Please note the logicle copyright information in the source files (package.html) and in the "Help-About" Menu.